Why is the modified wedge-opening-loaded test inadequate for characterizing gaseous hydrogen embrittlement in pipeline steels? A review

4.1 Selecting the initial K_Iapp

The initial applied stress intensity factor, K_Iapp, in a crack arrest test should be K_TH < K_Iapp < K_IC, Figure 2. Considering that the K_TH is obtained at the end of the test, sound engineering judgement and previous knowledge of similar materials are required to estimate K_Iapp. Point KD-1045 in ASME BPVC VIII-3 (2017) states that K_Iapp should be 1.5 times greater than the estimated K_TH but less than 198 MPa m^1/2 (180 ksi.in^1/2). The code provides a table with a range of recommended K_Iapp values for steels with yield strength between 621 MPa and 897 MPa (90 ksi and 130 ksi), Table 2.

The rationale behind Table 2 is a well-accepted result of hydrogen embrittlement that dictates that, for a given steel microstructure, K_TH increases with a decrease in yield strength (Bandyopadhyay et al. 1983; Echaniz et al. 1998; San Marchi and Somerday 2012; Yoshino and Minozaki 1986), hence, K_Iapp should increase as well. NACE TM0177 (2016) standard for measuring the stress intensity factor for crack arrest in hydrogen sulfide environments follows the same rationale, by imposing a larger CMOD with a decreasing yield strength. Extrapolating from Table 2, the K_Iapp for pipeline steels with an SMYS that is limited between 30 ksi (210 MPa) and 80 ksi (555 MPa) for hydrogen service (ASME B31.12 2023) should be well above 150 ksi.in^1/2 (165 MPa m^1/2). However, increasing K_Iapp is challenging in modified WOL tests because specimen dimensions increase with the square of K_Iapp, as will be shown below.

Point KD-1047 (ASME BPVC VIII-3 2017) allows setting K_TH = K_Iapp/2 when there is no measurable crack propagation after the 1,000 h exposure time. It is interpreted that dividing the K_Iapp by two might yield a conservative K_TH value, but the technical basis for this is not explained in the code. Taking into account that K_TH = 55 MPa m^1/2 is the minimum required value for base metal and welds according to option B of ASME B31.12, a pragmatic solution is setting K_Iapp = 110 MPa m^1/2. If the crack does not propagate at this K_Iapp level, the pipeline would still be qualified for hydrogen service and the same safety coefficients as for natural gas service could be used per ASME B31.12 (2023) code. Furthermore, obtaining a valid K_TH from a constant displacement test where the crack did not propagate is not considered in ASTM E1681-03 (2020).

A possible argument for setting K_TH = K_Iapp/2 when the crack does not propagate could arise from an analysis of results by Nibur et al. (2013) on pressure vessels steels, Table 3. The table shows K_TH measured according to ASTM E1681 and the maximum value of K_Iapp that resulted in no crack propagation. In all cases, K_TH > 50 % max. K_Iapp for no crack propagation. However, extrapolating such result to pipeline steels exposed to gaseous hydrogen is unwarranted, especially considering that for such steels there are not reliable measurements of crack arrest values measured according to ASTM E1681. Notice in Table 3 that as the yield strength of the steel decreases, the maximum K_Iapp for no crack propagation increases, which also provides a basis to argue that K_Iapp for pipeline steels should be well above 144 MPa m^1/2.

The following sections will show that even if K_Iapp is set to 110 MPa m^1/2, for common geometries and mechanical properties of pipelines it is impossible to extract a standard modified WOL specimen that fulfills all ASTM E1681 restrictions.

4.2 Specimen thickness and side grooves

For air tests, the fracture toughness decreases with an increase in the thickness of the specimen (B) under study (Anderson 2005), eventually reaching a plateau. The thickness independent fracture toughness has traditionally been named plane strain fracture toughness (K_IC). The following requirement is indicative of plane strain conditions along the crack front (Joyce and Tregoning 2000), so that a conservative K_IC value is obtained:

Similar considerations are held for the environment affected fracture toughness test (ASTM E1681), where K_IC should be replaced by K_TH.

The thickness effect on measured fracture toughness is a consequence of the relative contributions of the shear fracture near the specimen sides and the flat fracture on the specimen center, where plane strain conditions prevail (Anderson 2005). The use of side-grooved specimens creates plane strain conditions along the entire crack front and could eliminate the shear fracture contribution (Nibur et al. 2013), even when the requirement of Equation (3) cannot be satisfied.

Besides eliminating shear lips, side grooves could help to guide the crack along a single plane and to produce a straight fronted crack (ASTM E1681-03 2020). The ASTM E1681 standard allows up to a 20 % total reduction in thickness by side grooves. Although Equation (1) is independent of thickness, the presence of side grooves affects the estimated value of applied and arrest stress intensity factors. Equation (1) was obtained by combining expressions of K_I (as a function of applied load and geometry) and of compliance, both of which have different dependences on total and reduced thickness. This generated a correction factor that for 16 % side grooves increases the K value about 4 % relative to the value determined if the side grooves were ignored (Nibur et al. 2010).

In the original paper by Novak and Rolfe (1969) 1 inch (25.4 mm) thickness specimens of maraging steels (σ _y = 1,276 (185) and 1,310 (190) MPa (ksi)) were loaded to around 90 ksi.in^1/2 and, exposed to synthetic seawater. After propagation of the crack and arrest, K_ISCC was 43.6 ksi.in^1/2 and 52 ksi.in^1/2. Hence, the requirement of plane strain condition in Equation (3) was fulfilled from the beginning to the end of the test. Semicircular side grooves were optional in the modified WOL geometry proposed by Novak and Rolfe. They had a depth equal to 5 % of the specimen thickness. In all cases, after the test ended, examination of fracture surfaces revealed that a flat fracture surface propagated in the environment. Loginow and Phelps (1975) also used 5 % side grooves for studying K_TH of various pressure vessel steels exposed to pure hydrogen. The thickness was also 1 inch, and plane strain conditions prevailed at crack arrest. The crack front was planar and essentially straight. For different pressure vessel steels studied by Nibur et al. (2013), that did not fulfill Equation (3) requirement, 16 % side grooves eliminated shear lips along the entire crack length, suggesting that plane strain conditions dominated.

For pipeline steels, using 50 ksi.in^1/2 for the minimum K_TH admitted by the ASME B31.12 code, and considering the range of possible SMYS values, Equation (3) requires a minimum B of about 176 mm (7 inches) for an API 5L grade A pipeline and 25 mm (1 inch) for an X80 pipeline, well above the wall thickness commonly found in pipelines. The KD-10 article (ASME BPVC VIII.3 2017) allows to set the thickness of the modified WOL specimen equal to 85 % of the wall thickness of the component under study. When the requirement of Equation (3) is not met, ASTM E1681 indicates that K_TH is thickness dependent (in the standard K_IEAC and K_EAC are used to refer to plane strain and thickness dependent parameters, respectively, Table 1). The usefulness of a thickness dependent parameter for integrity evaluations (like allowable pressure or critical crack size, Kappes and Perez 2023b) is questionable, even when the thickness of the modified WOL specimen is similar to that of the component under evaluation (Anderson 2005). The reason is that depending on component and crack geometry and orientation, the size of the high triaxiality zone ahead of the crack might not depend on the specimen thickness, Figure 3. A K_TH value obtained with the specimen on the left, even when B = 0.85 t, could not be adequate to estimate the stability of the crack in the structural component showed on the right. The relative contribution of shear fracture in the test specimen is larger than in the structural component, even when B = 85 % t.

Figure 3:

Comparison of zones with high and low triaxiality ahead of a crack in a fracture mechanics test specimen with thickness B (left), and ahead of a surface crack in a component of thickness t (right). Notice that in the right, the stress state and hence fracture morphology might not be directly related to specimen thickness. Adapted from Anderson (2005).

4.3 Specimen width (W)

According to ASTM E1681, a valid K_TH requires that the remaining ligament (W−a) fulfills Equation (4).

This ensures that the plastic zone is sufficiently small so that the fracture is under LEFM regime. Notice that in ASTM E1681, the above condition is placed using K_TH, not K_Iapp. However, taking into account that KD-10 article (ASME BPVC VIII.3 2017) considers that a valid test can be obtained even when the fracture did not propagate, the relation as shown in Equation (4) provides the most conservative condition. A valid test requires a higher W as the a/W ratio increases and the yield strength decreases. Using 110 MPa m^1/2 for K_Iapp, Equation (4) requires a minimum W−a of about 359 mm (14 inches) for an API 5L grade A pipeline and 50 mm (2 inches) for an X80 pipeline. Figure 4 shows minimum W values for different API 5L pipeline grades and different a/W ratios. For a better visualization of the data, the figure is restricted to grades with SMYS equal to or greater than 42 ksi, which are the most common in existing pipelines.

Figure 4:

Minimum W required by ASTM E1681-03 (2020) for different API 5L pipeline grades and a/W ratios. It is considered that for each grade σ _y  = SMYS.

4.4 Specimen height (2H)

In the standard configuration (ASTM E1681-03 2020, ISO 7539-6 2018), the modified WOL specimen has an H/W ratio of 0.486. For specimens cut in the TL direction, the curvature of the pipeline might preclude fulfilling this condition on H, Figure 5. Considering that the specimen width must be at least 85 % of the wall thickness (t), the following relationship holds between the maximum possible value of H and the pipeline external diameter (D_ext):

Figure 5:

Extraction of a modified WOL specimen in the TL direction from a pipeline segment, showing standard height and B equal to 85 % of t.

Solving for the required external diameter D_ext,req when H/W = 0.486:

Figure 6 shows D_ext,req for obtaining WOL specimens of the standard geometry. D_ext,req rises with an increase in a/W and a decrease in t. The increase in the required D_ext with a decrease in pipeline yield strength is explained considering Figure 4.

Figure 6:

Dependence of minimum external diameter of pipeline required to obtain standard ASTM E1681 modified WOL specimens (H/W = 0.486), for different values of pipeline wall thickness (t), a/W ratios and pipeline yield strengths.

For the range of diameters typically (Folga 2007) found in transmission pipelines (shaded in gray in Figure 5), specimens with a/W = 0.5 can only be obtained when the actual σ _y is above 80 ksi. A modified WOL specimen made from a pipeline with σ _y = 65 ksi or below and with a wall thickness up to 0.5 in (12.7 mm) requires a diameter higher than the maximum pipe diameter typically found in transmission pipelines. For existing natural gas transmission pipelines with actual σ _y between 42 and 80 ksi and wall thickness up to 0.5 in (12.7 mm), it is impossible to obtain specimens with the conventional H/W ratio and a/W = 0.5 or a/W = 0.7. Whenever specimens with H/W = 0.486 cannot be obtained from the parent product, the investigator must procure the corresponding coefficients in Equations (1) and (2) by numerical modelling, adding costs and time to the project.

The limitation depicted in Figure 5 applies also for obtaining specimens in the LT direction, required for characterizing circumferential welds and HAZ. In the LT direction, the modified WOL specimen could have an unlimited value of H, but W will be limited by pipeline curvature. Considering that 2H < W (Figure 1), obtaining specimens in the LT direction is even more stringent than in the TL direction. The use of hot or cold flattening of the pipeline is not acceptable because the steel microstructure and the fracture mechanical properties would be affected.

6.1 Review of results obtained on pipeline steels at constant displacement

Table 4 summarizes literature results of constant displacement tests in gaseous hydrogen obtained on steels with mechanical strength within that of pipeline steels. Tazedakis et al. (2021) studied API 5L grades X60, X65 and X70, with actual σ _y between 492 and 539 MPa (71–78 ksi) in 8 MPa H₂ and with an K_Iapp between 110 MPa m^1/2 and 144.6 MPa m^1/2. The authors reported no crack propagation at the end of the 1,000 h testing period. The specimens were sized following the recommended configuration in ASTM E1681, W = 2B. However, all the specimens had a W-a value lower than the ASTM E1681 requirement in Equation (4), so the test is invalid according to KD-10 article in ASME BPVC VIII-3 (2017).

Cravero et al. (2023) studied an API 5L X65 (with actual σ _y = 501 MPa = 72.7 ksi) in 40 MPa H₂ at K_Iapp between 117 MPa m^1/2 and 187 MPa m^1/2. Similarly to Tazedakis et al. work, W was set equal to 2B. The pipe had a wall thickness of 19 mm, which is in the upper range of wall thickness typically encountered in existing pipelines. The authors acknowledge that (W−a) was smaller than required per ASTM E1681. The smaller specimen size was validated numerically, showing that the difference between K_Iapp calculated from an elastic-plastic finite element model and the prediction from linear elastic solution (Equation (1) and (2)) was less than 5 % for K_Iapp up to 112 MPa m^1/2. Incidentally, an increment in the initial crack size increased the maximum K_Iapp admitted according to the finite element modelling validation. This is also beneficial because an increase in a/W decreases the load in the bolt. Finite element modelling was also applied to reduce the conservatism of Equation (4) by Nibur et al. (2010). In one of the tests on a pressure vessel steel with σ _y = 717 MPa, the crack extended up to a/W = 0.81, so that the remaining ligament was below the minimum required by Equation (4). However, the differences between K_TH calculated with Equation (1) and (2), a finite element analysis assuming small scale yielding, and an elastic plastic analysis were all within a 10 % error. Furthermore, comparing results from a small scale yielding versus an elastic plastic model, there was less than 2 % difference in the magnitude of opening stresses on the plane of symmetry ahead of the crack.

Loginow and Phelps (1975) studied several pressure vessel steels in H₂ at a pressure ranging from 21 to 97 MPa. The tested steels with lower strength had a σ _y between 290 MPa and 586 MPa (42 ksi to 85 ksi), which is in the range of actual σ _y of pipeline steels admitted by ASME B31.12 code. In none of those steels, the authors observed crack propagation. The specimens had a thickness of 25.4 mm and were loaded to K_Iapp ranging from 55 to 116 MPa m^1/2. While the applied K_Iapp was in some cases low considering current requirements of testing standards, it is acknowledged that this paper preceded the first editions of ASTM E1681-03 (2020) and ASME BPVC VIII-3 (2017). Loginow and Phelps also tested pressure vessel steels with σ _y between 586 MPa and 1,055 MPa (85 ksi–153 ksi), which exhibited crack propagation and arrest in pressurized hydrogen. However, those results are not included in the table because their strength is well above that of pipeline steels.

Subcritical crack growth was not detected in constant displacement tests performed in compact specimens (Cialone and Holbrook 1988) in gaseous hydrogen and mixtures, in base metal and heat affected zones of X70 and X42 pipeline steels, respectively.

Two papers (Cialone and Holbrook 1988; Xu 2012) reported crack propagation in pipeline steels exposed to gaseous hydrogen in constant displacement tests. In both studies, the metal had an unusually high hardness (>38 HRC) and susceptible microstructure, resultant from a quenching process or fast cooling in the HAZ of an inappropriate welding procedure.

Considering the stringent restrictions detailed in the previous sections and the high load in the bolt, in all studies on base metal of linepipe steels found by the authors of the present paper, at least one of the requirements imposed by the testing standards was not fulfilled. The lack of valid tests reported in literature is at odds with the fact that there is a wordwide interest in qualifying existing transmission pipelines for hydrogen gas transport. While finite element modelling could reduce some of the conservatism in sizing restrictions, it must be done case by case adding time and costs to experimental programs. Additionally, crack propagated only in regions of unusually high hardness or susceptible microstructure. In this regard, ASME B31.12 (2023) places a restriction of 235 HV (Vickers hardness) on base metal, weld and HAZ. This hardness is less than 20 HRC (Rockwell hardness in scale C). Hence, the cases where crack propagated in Table 4 had an unacceptable hardness according to ASME B31.12.

6.2 Review of results obtained on pipeline steels following ASTM E1820 methodology

Although ASTM E1820-20b (2020) is not currently considered by ASME B31.12 for characterization of hydrogen affected fracture toughness, literature results obtained with this testing procedure are useful for getting some conclusions about material performance in constant displacement tests. For instance, it is accepted (Al-Rumaih and Gangloff 2005; Nibur et al. 2013) that for steels with σ _y < 900 MPa, the K_IH obtained under slowly raising CMOD provides a lower bound estimate of the value to be obtained in constant CMOD tests (ASTM E1681-03 2020). Even though ASTM E1820 is a standard originally developed for air tests of fracture toughness, its procedure for obtaining the onset of crack extension from J or CTOD measurements is extrapolated to tests performed in a pressurized autoclave with H₂. Most authors calculate K_IH from J_IC obtained using the 0.2 mm offset curve traced parallel to the blunting line curve according to ASTM E1820. Notice that K_IH describes resistance to crack initiation plus a small amount of crack extension (Nibur et al. 2010). K_IH is obtained from J_IH and CTOD_H according to (ASME BPVC VIII-3 2017), using Equations (17) and (18).

where the subscript H indicates that the measurements are conducted in pressurized hydrogen and  ν is Poisson ratio. The onset of crack extension can be estimated with the direct current potential drop (DCPD) technique. DCPD monitors crack length by applying a constant current through the specimen and measuring the potential drop at the crack mouth, an increase in the potential drop can be related to an increase in crack length (ASTM E1820-20b 2020). However, the actual value for the onset of crack extension is very sensitive to the interpretation of the DCPD data, hence, the 0.2 mm offset construction of ASTM E1820 for tests in air is widely adopted.

Table 5 is a non-exhaustive list summarizing measurements of K_IH obtained from J integral or CTOD tests normalized in ASTM E1820, performed in pressurized pure hydrogen in base metal of pipeline steels. For most reported tests, K_IH is greater than 100 MPa m^1/2. Similar results were recently reported by Steiner et al. (2023). K_IH of steels from many pipeline grades used in German natural gas transmission network were analyzed in 10 MPa H₂ and in most cases K_IH was also >100 MPa m^1/2 for the base metal.

The sizing requirements in ASTM E1820 are.

where ν = 0.3 was adopted for ferritic steels. Considering the respective SMYS for pipeline steels, σ _y/E ranges from 1.4 × 10⁻³ for an API 5L X42 up to 2.8 × 10⁻³ for an API 5L X80. Hence, the sizing requirements of ASTM E1820 are much more lenient than those of ASTM E1681 (Equations (3) and (4)). For example, for K_IH = 164 MPa m^1/2 (largest value in Table 5) and σ _y = 290 MPa (SMYS of an API 5L X42), the thickness B and initial uncracked ligament should be greater than 4.2 mm. A specimen with such dimensions can be easily extracted from most high-pressure transmission pipelines. However, validity of the test according to ASTM E1820 standard also requires straightness of the crack front. In this regard, crack tunneling and delamination have been reported after rising displacement tests of pipeline steels in gaseous hydrogen (San Marchi et al. 2010, 2011), which can prevent meeting the requirement for straightness of crack propagation.

For rising displacement tests in gaseous hydrogen, a key experimental parameter that has a direct effect on the measured K_IH value and J curve is the load displacement rate. A too rapid rate can grow the crack at a faster rate that hydrogen can diffuse ahead of it and towards the region of high triaxial stresses near the crack tip. Tests performed by Clark and Landes (1976) on a type 4340 Ni–Cr–Mo steel with σ _y = 1,240 MPa in 0.55 MPa H₂ showed that K_IH decreased from 77 to 30 MPa m^1/2 when the loading rate decreased from 0.8 MPa m^1/2/s to 2 × 10⁻³ MPa m^1/2/s. At the low loading rate, the K_IH value measured on this high strength steel was equivalent to the K_TH measured at constant displacement, Figure 9. Similar results were reported by the author at a σ _y = 1,102 MPa. Bezensek et al. (2023) tested the base metal of API X60 steel in 1.55 MPa H₂, at different loading rates ranging from 1.7 × 10⁻² MPa m^1/2/s down to 3.3 × 10⁻⁵ MPa m^1/2/s. In this range of K rate, K_IH decreased from 132 to 111 MPa m^1/2, Table 5. In addition to the effect on K_IH, the displacement rate decreased the slope of the J curve (Bezensek et al. 2023). Hence, besides the effect on K_IH, an increment in the K rate increased the difference between J_IH at the 0.2 mm offset line and the corresponding value measured at the onset of crack growth. For the smallest K rate of 3.3 × 10⁻⁵ MPa m^1/2/s, both parameters were similar. However, guidelines on proper loading rates for pipelines are still under development (EPRG 2024), and it is currently recommended to explore the effect of loading rate at the beginning of a testing program. For instance, similar K_IH values were measured at 0.05 and 0.005 MPa m^1/2/s in tests by San Marchi et al. (2010). Cialone and Holbrook (1988) did not observe differences in K_IH of API 5L X42 and X70 in 6.9 MPa H₂ measured at a displacement rate of 2.5 × 10⁻⁶ and 2.5 × 10⁻⁷ m/s, Table 5. The K rate during elastic loading is controlled by load displacement rate, specimen geometry and dimensions and machine compliance.

Figure 9:

Comparison of K_TH and K_IH for pressure vessel steels in pressurized gaseous hydrogen, after Nibur et al. (2013) and Clark and Landes (1976).

6.3 The relationship between K_IH measured with ASTM E1820 and K_TH measured with ASTM E1681

K_IH measured under rising CMOD and K_TH measured under constant CMOD represent critical stress intensity values for different processes (Nibur et al. 2013). In a rising CMOD test (ASTM E1820-20b 2020), hydrogen absorption and plastic straining at the crack tip are simultaneous and K_IH is the K_I value required for measurable crack advance. In a constant CMOD test, first the specimen is mechanically loaded in an inert environment, which generates a plastic deformation gradient near the crack tip. Afterwards, the loaded specimen is exposed to the environment where atomic hydrogen is absorbed at the crack tip and is transported to the fracture process zone. The crack propagates and is arrested at K_TH. Plastic strain in metals implies dislocation movement, and dissolved hydrogen enhances this movement (Robertson et al. 2015). As fracture is assisted by localized plastic strain, the different plastic deformation histories near the crack tip in both tests partly explain differences in critical stress intensity parameters. Additionally, for elastoplastic materials that require a critical amount of crack tip strain for cracking, like pipelines, different values of K_IH and K_TH are predicted based on differences in the strain field ahead of a stationary versus propagating crack (Nibur et al. 2013). Those are possible reasons for the fact that, for steels with σ _y < 900 MPa, K_IH < K_TH (Al-Rumaih and Gangloff 2005; Gangloff 2003; Nibur et al. 2013), Figure 9.

As reported in item 4.1, the minimum K_Iapp for qualifying a steel for hydrogen service is 110 MPa m^1/2. Comparison of K_Iapp with K_IH in Table 5, indicates that:

Notice that pipeline steels for hydrogen service have an actual σ _y that ranges between 210 MPa and 705 MPa per ASME B31.12 (2023) code and API 5L (2018). Extrapolating the result in Figure 9, obtained on pressure vessels steels, to pipeline steels:

Hence, even if a standard modified WOL specimen could be extracted from the base metal of a pipeline steel, the expected outcome of the test would be “no crack propagation” at K_Iapp = 110 MPa m^1/2, unless the region ahead the crack tip has a highly susceptible microstructure or an unusually high hardness, Table 4. A susceptible microstructure or a high hardness region can be more effectively and economically evaluated by metallographic observation or microhardness testing. In this regard, a major limitation of fracture mechanics tests applied to the characterization of microstructural and chemical inhomogeneities in a material is the difficult alignment of the crack plane with the most susceptible zone in the material.

Figure 10 presents another challenge of measuring K_TH with a constant displacement test in a material that has a rising J_R curve (bold curve in Figure 10), like pipeline steels in pressurized hydrogen environment (Bezensek et al. 2023; Cialone and Holbrook 1988; San Marchi et al. 2010). This challenge arises purely from mechanical effects (Anderson 2005). CMOD curves represent the crack driving force J at constant displacement, with CMOD_i < CMOD_i+1. The crack driving force decreases at constant displacement as the crack advances. The intersection point of crack driving force with J_R curve defines the crack arrest point. The higher the initial crack driving force, the higher the ordinate of the intersection. Considering that K = J E / ( 1 − v 2 ) , based on the material performance in ASTM E1820 tests in pressurized hydrogen (Cialone and Holbrook 1988; San Marchi et al. 2010) an increasing value of K_TH is predicted with increasing K_Iapp. ASTM E1681 establishes that K_TH should be obtained with the lowest K_Iapp value where there is evidence of crack advance. An appendix of NACE TM1077 (2016) for double cantilever beam tests in hydrogen sulfide also recognizes that, for the range of K_Iapp values that yield a valid K_I for crack arrest (K_TSSC), K_TSSC decreases with decreasing K_Iapp. This appendix offers a procedure for obtaining a conservative K_TSSC value.

Figure 10:

Driving force (J) for crack propagation under different values of constant displacement (Δ_i < Δ_i + 1) and material resistance to crack advance (J_R), adapted from Anderson (2005).

6.4 On the risk of subcritical cracking at K_I below K_IH

For pipelines, although K_IH values measured under rising displacement conditions provide conservative results in comparison to modified WOL specimens, such values should not be interpreted as immunity “thresholds” for hydrogen stress cracking. While the concept of “thresholds” (ASTM E1681-03 2020) has extensive use in engineering practice, fundamental studies conducted in the last decades have questioned their validity. Various authors performed in situ crack growth rate tests, according to experimental procedures proposed by Andresen (Andresen 1988, 2019) and using high resolution systems capable of resolving crack propagation rates below 10⁻⁹ mm/s. Those tests showed that there is a sustained, measurable crack propagation rate below thresholds traditionally accepted for cracking of structural materials in high temperature water. Experimental procedures involve the transitioning of cyclic loading of a fatigue crack to sustained loading conditions, by gradual increasing the load ratio (R = K_min/K_max, where K_min and K_max represent minimum and maximum K_I applied during a stressing cycle) and the time spent at K_max. This testing technique proposed by Andresen (Andresen 1988, 2019) has recently expanded beyond the nuclear materials field and was used to study EAC of pipeline steels and corrosion resistant alloys used in oil and gas service (Sridhar et al. 2017; Thodla et al. 2020). Recently Bezensek (Bezensek et al. 2023) applied this methodology to a pipeline steel in gaseous hydrogen.

Bezensek et al. (2023) evaluated an API 5L X60 steel in 1.5 MPa H₂. It was first fatigue loaded, at ΔK of 35 MPa m^1/2 and ΔK values were decreased until 2.2 MPa m^1/2. Then the frequency was reduced from 0.1 Hz down to 1 mHz, thus loading was gradually transitioned to constant K_I. At this point, there was a measurable subcritical cracking rate at the K_I = 40 MPa m^1/2 level. This measurable subcritical crack growth rate contrasts with a K_IH above 100 MPa m^1/2 measured under slowly rising displacement (Table 5), hence confirming that K_IH should not be used as a threshold value for crack advance. Those results are in accord with the performance of similar materials in sour environments (Thodla et al. 2020), where the source of hydrogen is different but the H-material interactions at the crack tip responsible for hydrogen stress cracking should be similar.

Very low crack growth rates might be pragmatically interpreted as immunity (Andresen 2019), depending on the structure expected service life. The situation is similar to fatigue crack growth, where a ΔK_th fatigue crack growth threshold might be pragmatically defined as the ΔK value when the growth rate is 10⁻¹⁰ m/cycle (ASTM E647-15 2015). In this regard, it is interesting to compare the rate of subcritical crack growth of 5 × 10⁻⁸ mm/s at 40 MPa m^1/2 reported in (Bezensek et al. 2023) with usual testing times in constant displacement tests. For a 1,000 h test, this represents a crack advance of 0.18 mm. The value is below 0.25 mm, the minimum crack extension required for distinguishing “crack propagation” from “no crack propagation” according to ASME BPVC VIII.3 (2017). Previously reported results show that the current resolution of crack growth rate techniques questions the significance and interpretation of thresholds for EAC. Additionally, it can be concluded that the risk of subcritical cracking at K_I below K_IH is a topic that deserves further study.

6.5 The effect of oxides on hydrogen absorption

For a steel exposed to gaseous hydrogen, the rate of hydrogen absorption decreases by native oxide layers on the surface, formed during prior contact with air (Nelson 1974; Turnbull 2012; Yamabe et al. 2015b). Hydrogen absorption through an oxide covered surface involves equilibrium between adsorbed hydrogen on the oxide and absorbed hydrogen in the oxide and equilibrium between absorbed hydrogen in the oxide and absorbed hydrogen in the steel (Turnbull 2012). Hydrogen is more soluble in the oxide than in the metal but the diffusion coefficient through the oxide is much smaller (Yen 1999). This results in a charging flux of hydrogen (J_ch) through an oxide covered surface that is much smaller than through bare metal (Yen 1999), and in an effective decrease in the solubility of hydrogen in the underlying metal. Such effect has been observed in hydrogen permeation experiments (Yen 1999) and in experiments where cylindrical coupons of a low alloy steel were hydrogenated in 100 MPa H₂ at room temperature (Yamabe et al. 2015a). Those specimens did not reach the expected saturation concentration even after 300 h, although that time is more than 6 times higher than the expected saturation time based on a diffusion analysis (Yamabe et al. 2015a). Coating or sputtering the surfaces with palladium eliminates the retarding effect of oxides (Bruzzoni and Garavaglia 1992; Koren et al. 2023; Yamabe et al. 2015a). Alternatively, when structures or specimens are stressed in hydrogen, the oxide barriers break, exposing fresh metal to the gas phase and favoring equilibrium between H₂ and H in interstitial lattice sites. This is particularly true in the presence of elevated plastic strain at the tip of stressed cracks. It is possible that oxides can lead to unconservative estimations of K_TH if the WOL specimens are loaded in air and then tested in pressurized hydrogen during an insufficient testing time. The alternatives for solving this issue included an increase in testing time, loading the WOL specimen in the autoclave, loading the WOL specimen in a controlled atmosphere, then transferring then it to the autoclave, or coating the specimen with palladium.

Loginow and Phelps (1975) used the modified WOL specimen for studying K_TH of low alloy steels in gaseous H₂ at 21–97 MPa. The specimens were stressed in air and placed in the test vessel on the following day. The specimens were instrumented with a strain gage in the reaction pin, so that the time for the onset of crack propagation and arrest could be monitored. The incubation time was several thousand hours, but in some cases, it was a few hours. The incubation time could not be correlated to mechanical or metallurgical factors of the WOL specimens. The authors related it to the kinetics of hydrogen permeation through the oxide layer at the crack tip. Nibur et al. (2010, 2013) tested several pressure vessel steels at 103 MPa. In some cases, loading the WOL specimens in air resulted in incubation times longer than if the specimens were stressed in argon, but the final K_TH values were similar in both cases. There was a decrease in incubation time with an increase in K_Iapp, irrespective of whether the samples were loaded in air or in argon. The causes for the large dispersion in incubation time are still not solved.

The KD-10 article (ASME BPVC VIII-3 2017) sets that the minimum testing time should be at least 1,000 h. Additionally, the loading process must be performed in a glove box under a controlled atmosphere, to circumvent the oxidation of the crack tip (ASME BPVC VIII-3 2017). The oxides at the crack tip, formed during fatigue precracking in air are ruptured during specimen loading, and could not reform due to the low oxygen and water concentrations in the glove box (Somerday et al. 2007). The glove box inert atmosphere should contain <5 ppm O₂ and <50 ppm H₂O. The loaded specimens must be transferred to the autoclave located inside the glovebox and pressurized with pure hydrogen at the desired pressure. Using a glove box reduced the incubation time (Nibur et al. 2010, 2013) but did not eliminate it. Loading of the bolt in constant CMOD tests can also be performed while the specimen is in the testing autoclave (Oriani and Josephic 1972), although this is not a common choice.

Another option would be to coat the specimen with palladium, which is known to favor equilibrium between dissolved hydrogen in the lattice and pressurized hydrogen (Yamabe et al. 2015a, b). Fracture mechanics tests could then be loaded in air. In the pressurized autoclave, hydrogen would be preferentially absorbed from the specimen sides rather than from the crack tip, but migration of hydrogen to the fracture process zone would still be possible during the standard testing time (Turnbull 2015). This alternative is not currently considered in KD-10 article (ASME BPVC VIII-3 2017).

Oxides at the crack tip resultant of fatigue pre-cracking in air are not a problem in rising CMOD tests, where the increasing plastic strain should effectively expose fresh metal and favor hydrogen absorption at the crack tip. A main challenge in those tests is to minimize oxygen contamination from leaks in the sliding seals (Holbrook et al. 1982).